Skip to content

Commit a0dcd67

Browse files
Set test batch size to 1
1 parent 7d1350c commit a0dcd67

File tree

3 files changed

+10
-8
lines changed

3 files changed

+10
-8
lines changed

config.py

+1-1
Original file line numberDiff line numberDiff line change
@@ -6,7 +6,7 @@ def get_default_config(args):
66
config.SAVE_EVERY_EPOCHS = 1
77
config.PATIENCE = 10
88
config.BATCH_SIZE = 450
9-
config.TEST_BATCH_SIZE = 256
9+
config.TEST_BATCH_SIZE = 1
1010
config.READER_NUM_PARALLEL_BATCHES = 1
1111
config.SHUFFLE_BUFFER_SIZE = 10000
1212
config.CSV_BUFFER_SIZE = 100 * 1024 * 1024 # 100 MB

preprocess.sh

+5-4
Original file line numberDiff line numberDiff line change
@@ -21,10 +21,11 @@
2121
# recommended to use a multi-core machine for the preprocessing
2222
# step and set this value to the number of cores.
2323
# PYTHON - python3 interpreter alias.
24-
TRAIN_DIR=/data2/edinella/java-small-clean-seq-c/training
25-
VAL_DIR=/data2/edinella/java-small-clean-seq-c/validation
26-
TEST_DIR=/data2/edinella/java-small-clean-seq-c/test
27-
DATASET_NAME=java-small-clean-seq-c
24+
PREFIX=/data2/edinella/seq-lim-b/
25+
TRAIN_DIR=$PREFIX/training
26+
VAL_DIR=$PREFIX/validation
27+
TEST_DIR=$PREFIX/test
28+
DATASET_NAME=seq-lim-b
2829
MAX_DATA_CONTEXTS=1000
2930
MAX_CONTEXTS=200
3031
SUBTOKEN_VOCAB_SIZE=186277

train.sh

+4-3
Original file line numberDiff line numberDiff line change
@@ -5,9 +5,10 @@
55
# test_data: by default, points to the validation set, since this is the set that
66
# will be evaluated after each training iteration. If you wish to test
77
# on the final (held-out) test set, change 'val' to 'test'.
8-
type=java-small-clean-seq-c
9-
dataset_name=java-small-clean-seq-c
10-
data_dir=data/java-small-clean-seq-c
8+
9+
type=seq-lim-b
10+
dataset_name=seq-lim-b
11+
data_dir=data/seq-lim-b
1112
data=${data_dir}/${dataset_name}
1213
test_data=${data_dir}/${dataset_name}.val.c2s
1314
model_dir=models/${type}

0 commit comments

Comments
 (0)